THS3491: THS3491

Hi Amina,

Can you send me the simulation circuit to check if there are any errors? The actual part will be able to source the 420mA specified in the datasheet. The THS3491 is a current feedback amplifier, which is optimized for high output current (420mA) and high bandwidth (900MHz) at the expense of precision. It works a little differently than a normal voltage feedback amplifier, so make sure to read the datasheet for instructions. 

For 500mA flat, you can put two open loop buffers (BUF634A) in parallel and buffer the output of an OPA2810, or any high voltage precision amplifier with otherwise insufficient output current. 

Best regards,

Sean

The THS3491 has a higher output and input headroom requirement than a lot of parts, make sure you are not voltage clipping. 

The imbedded current boost circuit can work, but be careful of phase margin loss due to prop delay in the boost amp. I showed a better way to do that in a recent AudioXpress article if you are interested. 

I'm interested.  How do I find your AudioXpress article?  Is it on the TI website?  I have an application where I need to create a virtual ground with very low impedance at 1MHz.  The original design used a 1GHz voltage-input op amp, with a gain-of-two THS3491 stage afterwards for more current drive.  It is oscillating madly, not surprising with all that GBW.  Does TI have an app note or a forum thread that addresses loop stability of a wideband composite amplifier containing a THS3491 ?  I understand control theory, but I would like to use existing expert knowledge rather than "roll my own" solution.  One thing that I need is the closed-loop phase vs frequency of the THS3491 at a gain of +2.  The gain vs frequency curve in the spec sheet shows a very sharp rolloff at around 800MHz, which usually means a lot of phase lag due to multiple poles.  Do I need to Spice it to get that data?  Thanks.

I just installed TINA TI.  Is there a tutorial?  I don't see a model for the THS3491 in the built-in "Spice Macros" library .  Can you help?

I was kind of looking at that BUF634A ckt above and recalling those AX article series kind of started with that (as a sort of a cautionary tale) - you never really want to bandlimit the amp inside the loop, and, that BW adjust feature is not even supported in the A version TINA model - kind of chased my tail on that one for a bit, but moved off to the better solution then. I will look for what month those articles start in the "subscription" AX site. They of course build on a lot of prior work over on Planet Analog in "The Signal Sped Up" series, particularly this one, 

https://www.planetanalog.com/stability-issues-for-high-speed-amplifiers-introductory-background-and-improved-analysis-insight-5/

Also, your comment about the THS3491 800Mhz gain and phase crashing down - I suspect youi are looking at a measured closed loop response where you will also eventually hit a slew limiting (even for this 9000V/usec device) which is probably what you are describing - but without the figure # no way to tell. 

Hi Gene, you can download the THS3491 model on the part's webpage https://www.ti.com/product/THS3491#design-tools-simulation

You can find tutorial videos here: https://training.ti.com/tina-ti-tutorial

The thing to remember is that this is a current feedback amplifier, so you have to use a minimum feedback resistor for it to be stable in its own loop. Then, you need to put it into a lower bandwidth loop. I can help you stabilize the loop if you want to share a schematic.

Here you go Gene, Some treatments of composite amplifer structure to get much higher output current.

The 1st one works through how to make that BUF634A circuit stable, the 2nd is much more capable and flexible. 

If you really only need 0V output (you said virtual ground) the highest output fast CFA is the dual OPA2677 - it has a linear output current spec, but will in fact source or sink >1.5A if the output voltage is midscale, 

Very helpful, thanks.  Let's see if I can paste that gain-of-2 frequency response curve: 

That is a brick-wall cutoff above 700MHz.  So I suspect that there is a lot of phase lag even at 30MHz.  Hopefully the Spice model will give me the correct phase.

So yes, I think this is the SSBW where you do indeed hit a lot of closed loop poles above 700MHz. Not sure those are going to matter to you in the band of interest and if this is inside the loop, it is really the full loop gain that matters for output impedance over frequency. 

The model is very good, I am not sure it includes package delay but that can be added externally. 

Still not clear on your full application - tidbits in your text

1. Active filter?

2. Virtual ground? 

We need a good virtual ground at the inverting input of the composite op amp, because we want to hold the lower terminal of a capacitor to nearly 0 volts.  So we need a lot of loop gain at 1MHz.  And we need to sink about 200mA which is coming out of the capacitor.  That's what the THS3491 is for.  I suspect I will need to do dominant pole compensation at the first stage to make the loop stable.  But I need to know the phase of the THS to find the dominant pole frequency.

We also place a small resistor in series with the input capacitor, to cancel the feedback pole.